University of Tasmania
Browse

File(s) under permanent embargo

Study of predictive models for emissions in hydrogen assisted dual fuel internal combustion engine

thesis
posted on 2023-05-27, 18:11 authored by Nguyen, VTP
A detailed understanding of the exhaust gas emissions can forecast the state of the engine performance and the other detrimental health effects it can have on the general population. There is no doubt that exhaust gas emissions generated by various internal combustion engines can provide environmental implications. With modem trends in alternative fuels and their mix with the conventional gasoline, is yet another effort to reduce exhaust gas emissions and adhere to strict emissions requirements in automobiles. While a good understanding of the quantitative and qualitative trends are available in the literature, for petrol driven vehicles, little or no published evidence is available for hydrogen vehicles or dual fuel driven vehicles assisted with hydrogen. A good understanding of the near zero emissions and associated conversion technology, using hydrogen as fuel, has been in the domain of few automotive companies around the world. While hydrogen is recognised as a potential fuel of the future, little or no evidence is available in the public domain on the mechanical and electrical conversion technologies and associated emission data for better understanding of this emerging alternative fuel. Conventional engine management systems with their inherent ability to map a particular fuel needed to be modified with dual fuel injection and particular add-on modular tools to accommodate hydrogen injection. This work is aimed at converting a commercially available Kawasaki Ninja 600cc motorcycle engine to run on both hydrogen and petrol. In this thesis, a rigorous design for conversion to run on hydrogen is designed and built from first principles. The test rig development associated with the calculations for fuel flow rates and associated engine management systems are integral part of this overall systematic design. As part of this investigation, an innovative fuel injection system together with add-on injection system is developed. Using artificial neural networks, predictive models for various mixtures of hydrogen-petrol are developed to estimate emissions for various hydrogen-petrol mixtures. It is argued in this thesis that the accuracy of prediction for emissions can replace expensive gas emissions equipment so that the intelligent mathematical predictive tools can be used as virtual sensors. As part of this investigation a comprehensive range of engine operating conditions is tested using both petrol and hydrogen as fuel for various combinations. The predictive model as virtual sensors has shown that the predictive capability for emissions is close to ±10% for various combinations of hydrogen-petrol fuel mix. Exhaust emission performance showed significant reduction in oxides of nitrogen and no significant emissions of hydrocarbons, carbon dioxide and carbon monoxide with increasing percentages of hydrogen injection. This work is seen a step towards understanding the intricate hydrogen conversions, development of add-on electronic injection control units to accommodate hydrogen and neural network based predictive models, as virtual sensors, to estimate internal combustion performance.

History

Publication status

  • Unpublished

Rights statement

Copyright 2007 the author Thesis (PhD)--University of Tasmania, 2007. Includes bibliographical references. Ch. 1. Introduction -- Ch. 2. Literature survey -- Ch. 3. Artificial neural network as predictive models for various non-linear dynamic processes -- Ch. 4. Experimental test rig set-up and calibration procedures -- Ch. 5. Embedded add-on fuel injection system with virtual emission sensor -- Ch. 6. Predictive models for emissions in petrol-hydrogen dual fuel engine using neural network -- Ch. 7. Online appraisal of the system -- Ch. 8. Final concluding remarks and proposed future works

Repository Status

  • Restricted

Usage metrics

    Thesis collection

    Categories

    No categories selected

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC